A5100740130- Advanced X-ray and CT Imaging
- Advanced Neural Network Applications
- Advanced SAR Imaging Techniques
- Radiomics and Machine Learning in Medical Imaging
- Retinal Imaging and Analysis
- Human Pose and Action Recognition
- Face and Expression Recognition
- Radiation Dose and Imaging
- Glaucoma and retinal disorders
- Image and Video Stabilization
- Phonetics and Phonology Research
- Video Surveillance and Tracking Methods
- Gastric Cancer Management and Outcomes
- Digital Imaging in Medicine
- Underwater Acoustics Research
- Cholangiocarcinoma and Gallbladder Cancer Studies
- Automated Road and Building Extraction
- Speech Recognition and Synthesis
- Image and Signal Denoising Methods
- Hepatocellular Carcinoma Treatment and Prognosis
- Medical Imaging Techniques and Applications
- Video Analysis and Summarization
- Retinal and Optic Conditions
- Digital Imaging for Blood Diseases
- Anomaly Detection Techniques and Applications
Sir Run Run Shaw Hospital
2021-2025
Zhejiang University
2021-2025
Xi'an University of Science and Technology
2015-2024
Fudan University
2024
Zhongshan Hospital
2024
The University of Texas Southwestern Medical Center
2014
Peking University
2005
An improved blood vessel segmentation algorithm on the basis of traditional Frangi filtering and mathematical morphological method was proposed to solve low accuracy automatic fundus retinal images high complexity algorithms. First, a global enhanced image generated by using contrast-limited adaptive histogram equalization image. Hessian model constructed introducing scale equivalence factor eigenvector direction angle matrix into enhance vessels Next, noise interferences surrounding small...
Abstract Purpose To study and investigate the synergistic benefit of incorporating both conventional handcrafted learning‐based features in disease identification across a wide range clinical setups. Methods materials In this retrospective study, we collected 170, 150, 209, 137 patients with four different types associated objectives : Lymph node metastasis status gastric cancer (GC), 5‐year survival high‐grade osteosarcoma (HOS), early recurrence intrahepatic cholangiocarcinoma (ICC),...
The SAR system possesses the ability to carry out all-day and all-weather imaging, which is highly valuable in application of aircraft identification. However, identification from images still faces great challenges due speckle noise interference, multiscale problems, complex background interference. To solve these an efficient bidirectional path fusion attention network (EBMA-Net) proposed this paper. It employs connectivity fuse features with different scales perform accurate detection...
Water extraction from synthetic aperture radar (SAR) images has an important application value in wetland monitoring, flood etc. However, it still faces the problems of low generalization, weak ability detailed information, and suppression background noises. Therefore, a new framework, Multi-scale Attention Detailed Feature fusion Network (MADF-Net), is proposed this paper. It comprises encoder decoder. In encoder, ResNet101 used as solid backbone network to capture four feature levels at...
In previous studies, we developed a blind deblurring algorithm based on the edge-to-noise ratio (ENR) and EM to improve quality of spiral CT images. this work, apply Wiener filter in place timing performance. The works well patient studies. After fully automatic deblurring, conspicuity submillimeter features cochlea image is substantially improved. new ENR-Wiener approach significantly improves computation speed with comparable
Statistical iterative reconstruction (SIR) methods have shown remarkable gains over the conventional filtered backprojection (FBP) method in improving image quality for low-dose computed tomography (CT). They reconstruct CT images by maximizing/minimizing a cost function statistical sense, where usually consists of two terms: data-fidelity term modeling statistics measured data, and regularization reflecting prior information. The SIR plays critical role successful reconstruction, an...
In order to extract facial features effectively, a face recognition method is proposed in this article, which combined with gamma transform and gabor on the basis of traditional filter method. First all, images are processed transform, can eliminate effects light other nonlinear factors; then decompose by improved transform. After that, instead original 5 scale 8 orientations images. At last M larger PCA (principal component analysis) characteristic every Gabor feature matrix, get final...
Due to practical application demand of automatic recorded education system, it is necessary detect and recognize students' actions in the classrooms. This paper proposes a method combining with Zernike moment motion history image computing optical flow class students, which mainly identifying three hands up, stand up sit down. moment, good sampling strong ability resist noise, can describe shape image. Lucas-Kanade direction speed movement. Based on video database Student-Behaviors captured...
This paper presents a new method for identifying the abnormal action of solitary oldies which is based on video sequence. First, we use background subtraction and morphological filtering technology to extract moving human contour. Then, motion energy image (MEI) body target, followed by extracting Hu moments feature extracted. At last, classify identify using Bayesian classifier. Experiments demonstrate that proposed recognition simple practical. It achieves correct rate daily behavior more...